An unsolved problem in the application of range split-spectrum method to L-band ALOS2/PALSAR2

Ionospheric effects are known to be one of the serious noises in L-band InSAR. However, splitting range-spectrum approach can virtually perform dual frequency measurements, and allow us to separate dispersive ionospheric phase and other non-dispersive phases (Rosen et al., 2010; Brcic et al., 2019; Gomba et al., 2016). Using the split-spectrum approach, Furuya et al. (2017) isolated the ionospheric signals on mid-latitude sporadic-E episodes. Thus the correction method of ionopheric phases appears to be already established, but here we would like to report and share an unsolved problem in the separation of dispersive and non-dispersive phases in ALOS2/PALSAR2 InSAR data. At the time of writing this abstract, we have encountered these problems only in some of the PALSAR2 InSAR data over higher latitude regions such as Iceland and Siberia, but we have not yet fixed the problem, either.

The problem is the appearance of strong phase gradient along the azimuth axis after separating into dispersive and non-dispersive phases, despite the fact that higher- and lower-frequency unwrapped differential interferograms do not reveal any such anomalies. Over the portions of the strong phase gradient, the dispersive and non-dispersive phases are in inverse sense. Even in those anomalous interferograms, the other areas outside the strong phase gradient seem to be adequately separated into dispersive and non-dispersive phases.

When we perform the multiple aperture interferomery (MAI) to both higher- and lower-frequency SAR images, we have encountered strikingly different MAI data that are hard to interpret only through the differences of the central frequency. When we apply the same MAI analyses to the sporadic-E InSAR images in Furuya et al (2017), we do not observe such significant differences. It seems we need to perform some additional procedures to correct for those anomalies.